python - Alternatives to scipy.interpolate.griddata that don't hang on aligned points -
i have point dataset i'm trying interpolate on grid. these points aligned in grid fashion points missing see below: 
to complicate it, it's possible other input pointsets may not align on grid, i'm trying use scipy.interpolate.griddata interpolate these values onto regular grid. however, underlying grid aligns sampling rate of input point dataset , griddata hangs. according this question scipy.interpolate.griddata operates poorly if sample occurs across 3 points aligned happens me.
is there high performance alternative interpolation of point data onto regular grid in python?
my solution has been stochastically perturb x , y coordinates of each input point random factor between 1e-6 * grid_cell_size. choose 1e-6 rule of thumb of several orders of magnitude greater minimum delta difference in 32 bit float, yet small enough error introduced still shadowed error introduced interpolation scheme.
i'm still open other ideas, way can use stock interpolation functions provided scipy introducing small numerical error.
Comments
Post a Comment